Time-frequency-energy representation based real-time speech recognition

Author

Wu, Duanpei ; Gowdy, J.N.

Author_Institution

Dept. of Electr. & Comput. Eng., Clemson Univ., SC, USA

fYear

1993

fDate

4-7 Apr 1993

Firstpage

0.625

Abstract

The authors present an approach to isolated-word speech recognition which is characterized by two aspects: (1) nonlinear time normalization based on the gradients of short-time energy in a specific number of frequency bands, which retains the transient portions and ignores the steady-state portions of the speech signal in the frequency domain; and (2) real-time implementation due to low computational load. Simulation has shown that the correct rate of recognition was 99.5% for multiple speakers based on the TI-20 speech database. A very high accuracy for on-line recognition was also obtained

Keywords

signal representation; speech recognition; time-frequency analysis; TI-20 speech database; frequency bands; frequency domain; isolated-word speech recognition; nonlinear time normalization; on-line recognition; real-time implementation; real-time speech recognition; recognition rate; short-time energy gradients; simulation; speech signal; time-frequency-energy representation; Artificial neural networks; Data mining; Feature extraction; Filter bank; Frequency; Hidden Markov models; Speech processing; Speech recognition; Steady-state; Vocabulary;

fLanguage

English

Publisher

ieee

Conference_Titel

Southeastcon '93, Proceedings., IEEE

Conference_Location

Charlotte, NC

Print_ISBN

0-7803-1257-0

Type

conf

DOI

10.1109/SECON.1993.465740

Filename

465740